156 Int. J. Electronic Marketing and Retailing, Vol. 5, No. 2, 2012 Measuring e-service quality in the context of service sites Estela Fernández Sabiote* and Sergio Román Facultad de Economía y Empresa, Campus de Excelencia Internacional Regional “Campus Mare Nostrum”, University of Murcia, Campus Universitario de Espinardo, 30100 Espinardo-Murcia, Spain Email: estelafs@um.es Email: sroman@um.es *Corresponding author Abstract: To deliver superior service quality, managers must first understand how consumers perceive and evaluate online customer service. This is especially the case for companies offering intangible products. Therefore, the objective of this research is to develop a scale to measure customer perceived service quality in the context of service sites (e.g. insurance, retail banking and travel). In addition, contrary to most of previous service quality studies, formative instead of reflective indicators are used to conceptualise e-service quality. Our findings, from two samples of 193 and 199 online consumers lead to a short easy-to-administer scale with two versions depending if the customer has not experienced a service failure (14 items) or he/she has experienced such failure (17 items). In the former case, functionality, information, reliability/fulfilment and privacy/security explain e-service quality. In the latter case, an additional dimension named – customer recovery is included. Results show a different pattern when the customer has any problem. Keywords: scale development; website service quality; services websites; formative indicator. Reference to this paper should be made as follows: Fernández Sabiote, E. and Román, S. (2012) ‘Measuring e-service quality in the context of service sites’, Int. J. Electronic Marketing and Retailing, Vol. 5, No. 2, pp.156–172. Biographical notes: Estela Fernández Sabiote is an Assistant Professor of Marketing in the University of Murcia, Spain. She has been a Visiting Scholar in the University of Maryland and University of Maastricht. Her articles have appeared in the International Journal of Market Research, Journal of Psychology and Business and Electronic Commerce Research and Applications. Her research interests are focused on services marketing. Sergio Román is an Associate Professor of Marketing in the University of Murcia, Spain. His articles have appeared in the Journal of the Academy of Marketing Science, Industrial Marketing Management, Journal of Business Research, Journal of Business Ethics, Journal of Marketing Management, Electronic Commerce Research and Applications, European Journal of Marketing, International Marketing Review, Journal of Business & Copyright © 2012 Inderscience Enterprises Ltd. Measuring e-service quality in the context of service sites 157 Psychology, International Journal of Market Research and British Journal of Educational Psychology, among other journals. His research interests are focused on personal selling and sales management, business ethics and services marketing. 1 Introduction Service quality has been recognised as having the potential to deliver strategic benefits, such as improved customer retention rates, whilst also enhancing operational efficiency and profitability (Cronin, 2003). The service quality literature has developed several relevant theories and measurement scales in the brick and mortar context (e.g. Parasuraman et al., 1985; Dabholkar et al., 1996; Brady and Cronin, 2001). Recently, research efforts on the measurement of consumers’ service quality perceptions includes focusing on a new channel – the internet – and the services of firms operating in this new context. Academic research on electronic service quality has speeded up over the past few years since the work by Zeithaml et al. (2002). This stream of research has developed an important knowledge about e-service quality and has shown agreement about the multidimensionality nature of this construct (e.g. Gounaris and Dimitriadis, 2003; Wolfinbarger and Gilly, 2003; Yang et al., 2004; Bauer et al., 2006; Collier and Bienstock, 2006; Kim et al., 2006; Akinci et al., 2010). Nevertheless, the aforementioned studies have generally ignored an important research priority suggested by Parasuraman et al. (2005): “to examine the scales in the context of pure-service sites, make any necessary modifications, and assess the psychometric properties of the modified scales” (p.229). According to Rust and Lemon (2001), e-retailers that sell only tangible products do not represent the true nature of eservice since most people think of their service as no more than access to goods and order fulfilment (p.86). Therefore, the existing scales can only be applied to contexts where there is a physical delivery of goods. In other words, these measurement instruments do not take into account the uniqueness of services offered by websites. For instance, some of their items can not be applied in a service context (e.g. ‘This eretailer’s orders are protectively packaged when shipped’, ‘Damage rarely occurs during transportation of my order from this e-retailer’, ‘Products on the site are almost always in stock’ and ‘The items sent by the site are well packaged’). Importantly, compared with tangible products, services have distinctive characteristics such as intangibility and perishability, which presents a range of challenges for consumers. In particular, consumers are generally unable to make a decision about the quality of a service until they have purchased it. This increases perceived risks as compared to the purchase of a physical product (Parasuraman et al., 1985). Research conducted in the traditional retail service context has consistently shown that the face-to-face interaction with front line employees reduces risk perceptions (Brady and Cronin, 2001). However, in the internet context, services customers have to interact with a website. It is not surprising that online retailers and scholars alike are interested in understanding how to measure e-service quality in the context of service sites. 158 E. Fernández Sabiote and S. Román Therefore, the goal of this article is to develop and validate an instrument to measure e-service quality in the context of service sites (e.g. insurance, retail banking and education). In what follows, we review prior studies about the measurement of e-service quality and how they relate to the conceptualisation, dimensionality and measurement of e-service quality in the context of service sites. 2 Literature review As argued above, in response to the growing recognition that a valid instrument for measuring e-service quality is crucial, several studies have been conducted in this area. Despite these important efforts, only very recently Akinci et al. (2010) adapted Parasuraman et al.’s (2005) scale to a ‘pure’ service context. In particular, their research measures consumers’ perceptions of the electronic service quality offered by 13 banks in Turkey. This study represents an initial step in the process of taking into account the particular characteristics of services offered by online retailers. Yet, their measure can only be applied to internet banking, and future research that covers different e-service industries is encouraged by the authors (Akinci et al., 2010, p.238). Importantly, only a limited number of studies have conceptualised and measure eservice quality as a second-order construct (e.g. Parasuraman et al., 2005; Bauer et al., 2006; Collier and Bienstock, 2006; Collier and Bienstock, 2009). This is particularly relevant because consumers use a high-level of abstraction to evaluate overall e-service quality and e-service quality is a complex and multifaceted construct (Rowley, 2006). The advantage of developing a higher-order conceptualisation of e-service quality is that it is a more restrictive theoretical model. In addition, most of the existing scales have used reflective instead of formative indicators, yet Parasuraman et al. (2005) have even questioned their own study’s use of reflective indicators by stating that: “based on the model specification criteria discussed by Jarvis et al. (2003), it might be more appropriate to treat the fist-order dimensions as formative indicators of the second-order latent construct” (p.220). Despite this call, only a few researchers have defended the use of formative scales to measure e-service quality (e.g. Collier and Bienstock, 2006; Rossiter, 2007; Collier and Bienstock, 2009). Importantly, when the concept intended to measure is a summative judgment such as eservice quality, the use of reflective instead of formative indicators can lead to model misspecification and ultimately to biased results1 (MacKenzie et al., 2005). In addition, following Jarvis et al.’s (2003) criteria, indicators in reflective models should be interchangeable, but e-service quality components such as fulfilment and privacy are clearly unique, distinguishable and not interchangeable. Also, with reflective measures, all components should co-vary with one another (Jarvis et al., 2003); however, though some components of e-service quality may correlate with the other components, no theoretical reason establishes that all must do so. Recently, Collier and Bienstock (2006, 2009) have empirically tested the formative structure of e-service quality. Two main implications are derived from their work: (a) eservice quality might be better represented by formative instead of reflective indicators and (b) the authors acknowledge the exploratory nature of their work and recognise that further testing using formative indicators is needed with other samples. Finally, scales for questioning the service problems that consumers encounter and their perceptions regarding the solution of these problems (i.e. service recovery) did not Measuring e-service quality in the context of service sites 159 exist until Parasuraman et al.’s (2005) developed the E-RecS-QUAL. Yet, they called for future research because: “the E-RecS-QUAL Scale should be viewed as a preliminary scale because the small samples of customers with recovery-service experience at the sites used in later stages of scale testing did not permit a comprehensive psychometric assessment of that scale” (Parasuraman et al., 2005, p.229). Within the light of these evaluations, the current study tries to take a further step in measuring electronic service quality by developing a scale that: (a) is particularly adapted to the context of websites that offer services; (b) measures e-service quality as a higherorder construct; (c) uses formative instead of reflective indicators and (d) provides a comprehensive psychometric evaluation of the e-service recovery scale. 3 Conceptualisation of service quality for service websites Zeithaml et al.’s (2002) qualitative research showed how consumers expect websites to have several quality features. These features range from specific, concrete cues (e.g. tab structuring, search engines and one-click ordering) to more general perceptual attributes (e.g. perceived ease of finding what one is looking for and perceived transaction speed) to broader dimensions (e.g. ease of navigation in general and responsiveness to customer needs) to higher-order abstractions (e.g. overall perceived quality and value). This is why the concept of e-service quality is framed by Parasuraman et al. (2005) in the means-endchain approach. This approach deems that consumers retain product information in memory at multiple levels of abstraction. Following this theoretical framework, the present study’s domain of e-service quality will be constituted by website features associated with the core evaluative process, that is perceptual attributes and dimensions (concrete cues are antecedents that influence this process). In this sense, we found not only theoretical but also empirical studies that, in an online context, suggest that eservice quality is a higher-order construct (e.g. Barnes and Vidgen, 2001, Francis and White, 2002, Parasuraman et al., 2005; Yang et al., 2005; Bauer et al., 2006, Collier and Bienstock, 2006). In a traditional (offline) context, perceived service quality is defined as a global judgement or attitude relating to the superiority of a service relative to competing offerings (Parasuraman et al., 1988). This definition is used by some researchers that develop scales to measure service quality in the online channel (e.g. Yang et al., 2003; Yang et al., 2004). Zeithaml et al. (2000) provided a widely accepted definition of eservice quality as “the extent to which a website facilitates efficient and effective shopping, purchasing, and delivery of products and services” (p.11). Based on the comments of focus groups conducted during the qualitative analysis, we further developed Zeithaml et al.’s (2000) definition to particularly focus on the context of online service providers. As such, in this study e-service quality is defined as “the extent to which a website facilitates the consumer buying process, payment, service delivery, and post-service experience in a competent, secure, and effective manner”. Given the variety of approaches for measuring e-service quality, we suggest a transaction process-based framework in order to capture all relevant quality aspects of the virtual service transaction. This approach provides richer diagnostic information and managerial implications for improving service quality (Bauer et al., 2006). Collier and Bienstock (2006) used a three-process framework (process, outcome and recovery), which is consistent with the feedback obtained in our qualitative study. In the ‘process’ 160 E. Fernández Sabiote and S. Román phase customers search for information about the e-service offerings, during this phase functionality and information are key service quality elements. In the ‘outcome’ phase, the transaction is accomplished. In this phase, security, privacy and reliable service delivery have been identified as the most important service quality aspects (Parasuraman et al., 2005). And finally, the ‘recovery’ phase occurs after the purchase when the customer has had a problem. It covers the manner in which service failures (if they occur) are handled. As shown in Figure 1, based on practitioner and academic literature (e.g. Gounaris and Dimitriadis, 2003; Wolfinbarger and Gilly, 2003; Barnes and Vidgen, 2004; Yang et al., 2004; Parasuraman et al., 2005; Yang et al., 2005; Bauer et al., 2006; Collier and Bienstock, 2006), we view e-service quality as a second-order factor (i.e. an aggregate construct) that is collectively formed – and not reflected –by either four or five reflective first-order dimensions (i.e. functionality, information, reliability/fulfilment, privacy/security and customer recovery) depending upon whether service provision was initially successful or required follow-up. Figure 1 Dimensions of the e-service quality scale E-SERVICE QUALITY FUNCTIONALITY INFORMATION RELIABILITY/ FULFILLMENT PRIVACY/ SECURITY The first-order dimension, functionality, refers to the correct technical functioning of the site. It is concerned with the pragmatics of how a user perceives and interacts with a website. Therefore, it includes the different technical aspects of the website that improve the customer-website interaction, namely: aesthetics (the appearance of the site), speed (the site’s correct and quick answers to customer demands), ease (the site is simple to use, structured properly, and requires a minimum of information to be input by the customer) and system availability (the correct technical functioning of the site). Following existing research (Gounaris and Dimitriadis, 2003; Yang et al., 2005; Collier and Bienstock, 2006) the Information dimension refers to the usefulness of the content as well as the adequacy of the information provided in the website. This seems to be a key dimension, for instance McGovern and Norton (2002, p.13) argued that: “the key different between commerce and e-commerce is that commerce is selling with people and e-commerce is selling with content”. Reliability/fulfilment refers to the probability that the company performs the service accurately, that is: delivering what is ordered under the conditions agreed (Yang et al., 2004, Parasuraman et al., 2005). Privacy/security can be defined as the degree to which the customer believes the site is safe from intrusion and personal information is protected (e.g. Wolfinbarger and Gilly, Measuring e-service quality in the context of service sites 161 2003; Kim and Lee, 2009). Similar to prior research (e.g. Wolfinbarger and Gilly, 2003; Yang et al., 2004; Parasuraman et al., 2005), both privacy and security aspects are integrated in this unique dimension. We add one dimension – customer recovery to the aforementioned four dimensions that is particularly salient when a service failure takes place. Customer recovery refers to the effective handling of problems by the online service company. It includes not only the availability of assistance through telephone or online representatives, but also correct and quick responses through the site when the customer has experienced any trouble (Parasuraman et al., 2005; Collier and Bienstock, 2006). 4 Methodology 4.1 Development of initial set of scale items First, through a review of a large base of the relevant literature, up to 233 items were identified.2 Then, three judges, who were given our e-service quality definition, were asked to place each item in one of three categories: ‘clearly representative’, ‘somewhat representative’ and ‘clearly not representative’. Additionally, the judges were also encouraged to examine the items in terms of general applicability and redundancy. In order to fill the gaps that were left by previous research, we conducted in-depth interviews and focus group interviews (Thompson et al., 1989). Five in-depth interviews and two focus group interviews were conducted with convenience samples of online consumers (i.e. people with experience in online shopping) in order to (a) improve the set of items; (b) generate new items; (c) eliminate any redundant, ambiguous or poorlyworded items and (d) identify key dimensions for each phase of the transaction. Prior to the items presentation, focus group participants discussed the meaning of e-service quality and their experiences with e-service websites.3 In all, 43 scale items were generated from the literature and focus group interviews (six of them were developed during the focus groups). 5 Data collection and analyses 5.1 First study Data collection was conducted by e-mail. 1312 personalised e-mails were sent to members4 of the community of a Southeastern University. The unit of analysis in this study is the individual consumer who has purchased services at least once from a website in the last four months (e.g. travel, insurance and online course). This facilitated consumers’ evaluations of the online retailer’s website. The e-mail message described the research purpose and invited each subject to participate in the survey by filling in the attached e-questionnaire if they had purchased a service online in the last four months. Subjects were asked to complete the questionnaire corresponding to the website where they had made their last online purchase during this period. After the elimination of missing data, 193 observations remained in our database (14.71% effective response rate).5 The respondents were mainly male (60.4%), relatively young (46.6% were 162 E. Fernández Sabiote and S. Román between 18 and 33) and generally highly educated (86% with a college degree). This is consistent with data reported in previous studies about the demographics of online shoppers (Kau et al., 2003; Swinyard and Smith, 2003). All scales consisted of 5-point multi-item Likert questions, ranging from ‘1 = totally disagree’ to ‘5 = totally agree’. Global service quality was measured using Yang et al.’s (2005) scale (three items). This scale is used as a reflective measure to validate our scale. E-service quality was measured with 30 formative items. An issue of particular interest to formative indicators is that of multi-collinearity (see Diamantopoulos and Winklhofer, 2001). Intercorelations of formative indicators may have a direct effect on the stability of the indicator coefficients as they are based on a multiple regression. High multi-collinearity among formative indicators could make it difficult to determine the impact of each indicator on the latent construct. In our sample, multi-collinearity among the indicators was low. The maximum variance inflation factor (VIF) (3.6) came far below the common cut-off threshold of 10. The nature of formative measurement renders an internal consistency perspective inappropriate for assessing the suitability of indicators; indeed, “the best we can do… is to examine how well the index relates to measures of other variables” (Bagozzi, 1994, p.333). In other words, due to the formative structure of the construct e-service quality, internal consistency, convergent and discriminant validity are not relevant. Both the measurement model and the structural model were estimated by means of partial least squares6 (SmartPLS, Version 2.0). PLS makes no distributional assumptions. Therefore, a bootstrapping method was used to ascertain the stability and significance of the parameter estimates (cf. White et al., 2003). More specifically, the t-values were computed on the basis of 500 bootstrapping runs. Sixteen of the 30 items were eliminated during these analyses. Contrary to Covariance-Based Structural Equation Modelling (CBSEM) PLS path modelling does not report any kind of fit indices like TFI, RMSEA or CFI (since PLS makes no distributional assumptions for parameter estimation). The evaluation of PLS model is therefore based on prediction oriented measures that are nonparametric (Chin, 1998). The PLS structural model is evaluated by R2 of endogenous latent variable (Chin, 1998), Goodness of Fit index (GoF) (Tenenhaus et al., 2005) and by using the Stone–Geiser Q-square test for predictive relevance (Stone, 1974; Geiser, 1975). According to Chin (1998), R2 values of 0.67, 0.33 and 0.19 for endogenous latent variables are described as substantial, moderate and weak, respectively, (R2e-service quality = 0.58). Goodness-of-fit (GoF) (Tenenhaus et al., 2005) was employed to judge the overall fit of the model. GoF, which is the geometric mean of the average communality (outer measurement model) and the average R2 of endogenous latent variables represents an index for validating the PLS model globally, as looking for a compromise between the performance of the measurement and the structural model, respectively. GoF is normed between 0 and 1, where a higher value represents better path model estimations (GoF = 0.63). Finally, the Q-squares statistics measure the predictive relevance of the model by reproducing the observed values by the model itself and its parameter estimates. A Qsquare greater than 0 means that the model has predictive relevance, whereas Q-square less than 0 mean that the model lacks predictive relevance (Fornell and Cha, 1994). In PLS, two kinds of Q-squares statistics are estimated, that is, cross-validated communality and cross-validated redundancy. Both statistics are obtained through blindfolding procedure in PLS. The q2 values of 0.02, 0.15 and 0.35 signify small, medium and large Measuring e-service quality in the context of service sites 163 predictive relevance of certain latent variable, thus explaining the endogenous latent variable under evaluation. All the results provide initial support to conceptualise eservice quality (a) with formative instead of reflective indicators and (b) as a secondorder construct composed of four dimensions: functionality, information, reliability/fulfilment and privacy/security (see Figure 1). 5.2 Second study A survey instrument was administered to a sample of 199 consumers.7 A marketing research firm was hired to assist with the data collection. Respondents were approached randomly among individuals who passed the data collection points located on the pedestrian walkway in two major metropolitan cities (for a similar procedure see Frambach et al., 2007, p.30–31). Since we preferred a more random sample of the population, quota sampling was not used. Instead, every fifth person who passed each of the data collection point located on the pedestrian walkway was invited to participate in the study (see Sekaran, 2002; Keen et al., 2004 for a similar procedure). Screening questions were administered before the respondent was invited for an interview. Also, they were asked if they had made more than one purchase during this period. Then, they were asked if they had experienced a problem and to specifically respond to the questionnaire based on this experience. If they had no problems, subjects were asked to complete the questionnaire corresponding to the website where they had made their last online purchase during this period. An invitation only followed if the respondent proved to be eligible for the study (i.e. he/she should have purchased a service online in the last four months). The latter condition to facilitate consumers’ evaluations of the online retailer’s website. Then, subjects were taken to the company office locations (conveniently located in the metropolitan areas). The procedure was to let subjects browse the website where they made their last online shopping. After a certain period of time (a maximum of ten minutes), subjects were asked to complete the questionnaire corresponding to that site. 45% of the respondents were male, the mean age was 30 years, and 71% of the respondents had a college degree. Again, this data is consistent with the demographic characteristics of internet users mentioned earlier. Similar to Study 1, all scales consisted of 5-point multi-item Likert questions, ranging from ‘1 = totally disagree’ to ‘5 = totally agree’. Seven items to measure customer recovery dimension were included in the questionnaire (based on Parasuraman et al.’s 2005 e-service recovery quality scale). In order to assess the nomological validity of the e-service scale, satisfaction with the website and trust in the website were measured with three-item scales adapted from Anderson and Srinivasan (2003). In this sample (full sample), multi-collinearity among indicators did not seem to be a problem. Therefore all items were retained for additional analysis. Data were estimated with SmartPLS 2.0. The PLS approach applies a bootstrapping method (500 samples, sample size 100) to calculate the t-values (all of them were significant). Again, the PLS structural model is evaluated by R2 of endogenous latent variable (Chin, 1998), Goodness of Fit index (GoF) (Tenenhaus et al., 2005) and by using the Stone–Geiser Q-square test for predictive relevance (Stone, 1974; Geiser, 1975). The R2, which indicates the extent to which the formative measurement model covers a construct’s scope (Diamantopoulos, 2006) is 0.52, the Goodness of Fit index for the e-service quality scale is 0.62 and the q2 164 E. Fernández Sabiote and S. Román is 0.35. Therefore, the model provides acceptable coverage of the e-service quality construct. Table 1 presents the final estimation results pertaining to our conceptual framework. Based on the model’s performance statistics, it can be concluded that the proposed model has a good fit to the data. Model test results for the e-service quality scale Table 1 Relationship Functionality → E-service quality Information → E-service quality Reliability/fulfilment → E-service quality Privacy/security → E-service quality Study 1 N = 193 Study 2 N = 199 Study 2 Mimic model 0.38(5.68)a*** 0.34(4.85)*** 0.34 (4.40)*** 0.24 (3.81)*** 0.24(3.25)*** 0.24 (3.03)** 0.20 (3.52)*** 0.13 (1.95)* 0.13 (1.94)* 0.12 (2.46)** 0.20 (2.70)** 0.20 (2.83)** E-service quality → Satisfaction E-service quality → Trust Notes: 0.50 (8.62)*** 0.60 (10.22)*** a t-values; ***Significant at p < 0.001; **Significant at p < 0.01; *Significant at p < 0.05. A final approach to validation, focusing on nomological aspects, involves linking the index to other constructs with which it would be expected to be linked (i.e. antecedents and/or consequences). Based on theory and empirical findings, e-service quality is expected to positively affect satisfaction and trust (e.g. Hennig-Thurau and Klee, 1997; Montoya-Weiss et al., 2003; Harris and Goode, 2004). Therefore, we estimated this MIMIC model by means of PLS (final estimation results are shown in Table 1). The model explained 51% of the variance in e-service quality, 25% of the variance in satisfaction and 35% of variance in trust. Then, we analysed the second scale, which is salient only to customers who had nonroutine encounters, and as such has experienced any trouble during or after the purchase process, with the sites and contains seven additional items in one dimension: customer recovery (see Figure 2). That is, we analysed 89 questionnaires of the second study. The limited sample size is not a problem when using PLS (Cassel et al., 2000). The measurement model and the structural model were estimated by means of PLS. As a result, four of the seven items were eliminated following the same criteria than in the full sample (the rest of the e-service quality dimensions remained the same). The R2 is 0.63, the GoF is 0.70 and the q2 is 0.44. Additionally, a MIMIC model by means of PLS was estimated. In this model satisfaction and trust are direct consequences of e-service quality. The model explained 63% of the variance in e-service quality, 26% of the variance in satisfaction and 44% of variance in trust. In sum, results indicated that the model had an acceptable predictive relevance. Measuring e-service quality in the context of service sites 165 Dimensions of the e-service recovery quality scale Figure 2 E-SERVICE RECOVERY QUALITY CUSTOMER RECOVERY FUNCTIONALITY PRIVACY/ SECURITY INFORMATION RELIABILITY/ FULFILLMENT 5.3 Overall results As shown in Table 1, functionality was the dimension with the strongest impact on customers’ e-service quality evaluations, followed by information in both studies. In the first study, the third and fourth factors were reliability/fulfilment and privacy/security, respectively, while this order is inverted in the second study. MIMIC model results confirm the dimensions pattern obtained in the second study. Also, MIMIC analyses provided strong support for the positive influence of e-service quality on satisfaction (standardised coefficient = 0.50; p < .001) and trust (standardised coefficient = 0.60; p < .001). Model test results for the e-service recovery quality scale Table 2 Relationship Functionality → E-service quality Information → E-service quality Reliability/fulfilment → E-service quality Privacy/security → E-service quality Customer recovery → E-service quality Study 2 N = 89 Study 2 Mimic model a 0.34 (3.46) *** 0.34 (3.51)*** 0.25 (2.74)** 0.26 (2.81)** 0.11 (0.88) 0.10 (0.94) 0.26 (2.97)** 0.27 (2.97)** 0.05 (0.66) 0.05 (0.69) E-service quality → Satisfaction E-service quality → Trust Notes: a t-values; ***Significant at p < 0.001; **Significant at p < 0.01; *Significant at p < 0.05. 0.51 (6.66)*** 0.67 (10.50)*** 166 E. Fernández Sabiote and S. Román Interestingly, results for the second scale, which is salient only to customers who had problems with the e-service (n = 89), showed a different pattern (see Table 2). Functionality had the strongest impact on customers’ e-service quality evaluations. The second dimension was privacy/security closely followed by information. Importantly, reliability/fulfilment and customer recovery did not have a significant effect on e-service quality evaluations when customers have experienced any trouble. These results will be discussed in the following section. Overall, the analyses (measurement model, structural model and MIMIC model) yielded a scale composed of 14 items (routine e-service quality scale) and three additional items to be added in case the customer has a non-routine encounter to yield 17 items in total. 6 Discussion This study represents the first effort to develop a general scale to measure e-service quality in the context of service sites using formative, instead of reflective indicators. This adds to the existing literature in two ways. First, items of our measure do particularly refer to service sites, and consequently they take into account the unique characteristics of services (intangibility, heterogeneity, inseparability and perishability). Second, e-service quality is measured with formative, instead of reflective indicators, which is consistent with Jarvis and colleagues’ criteria (Jarvis et al., 2003, Mackenzie et al., 2005). Numerous studies have conceptualised service quality as an attitude that is based on a reflective judgment. Conversely, drawing on recent research (e.g. Collier and Bienstock, 2006; Rossiter, 2007; Collier and Bienstock, 2009) following Jarvis and colleagues’ criteria, we believe that e-service quality is a summative judgment that takes place from evaluating numerous dimensions. In other words, the construct of e-service quality does not cause functionality or information. It is just the opposite; the dimensions of functionality, information, reliability/fulfilment and privacy/security form the overall evaluation of how the customer judges service quality in an online setting. Accordingly, this study adds further empirical evidence to the recent work of Collier and Bienstock (2006, 2009). The above implies that some of our results are consistent to previous reflective scales developed both for goods and for example, privacy/security and reliability/fulfilment are dimensions found in most of these studies (e.g. Wolfinbarger and Gilly, 2003; Parasuraman et al., 2005; Kim et al., 2006). Yet, our study presents new evidence. In particular, information proves to be a key dimension, which has the strongest impact on e-service quality. In this sense, McGovern and Norton (2002) compare having poor quality contents on the web with “making a customer in an offline store wait for ages for an assistant, only to find that the assistant gives them the wrong information” (p.27). In a service context, due to the intangibility and heterogeneity of services, customers are more likely to give more value to the quality of the information (i.e. ‘the content of the information is helpful and relevant’). Another key contribution of this study is the development and assessment of the eservice recovery scale. Instead of selecting the most frequently visited services websites, which is critical for sampling efficiency (Parasuraman et al., 2005) a more diverse context of services website has been used to examine this scale (Parasuraman et al., 2005 used two websites). This allows: (a) a higher incidence of problem encounters and (b) a Measuring e-service quality in the context of service sites 167 more realistic view of service websites performance since we are not focus on highly reputed sites. Importantly, the service recovery dimension places special emphasis on the attention given to customers when problems arise as well as how the problems are solved. The services brick and mortar literature has long evidenced that service recovery efforts have a major impact on customers’ positive evaluations of the service and the company (e.g. Maxham and Netemeyer, 2002; Maxham and Netemeyer, 2003). However, our data seem to indicate that online retailers’ actual recovery efforts have a non-significant effect on e-service quality. In other words, e-retailers actions to solve and compensate customers’ problems are not considered by customers as effective and satisfactory. In fact, Collier and Bienstock (2006, 2009) found that that service recovery failed to have a direct relationship with behavioural intentions. In addition, it is reasonable that reliability/fulfilment perceptions (e.g. the delivered service corresponds to the website offerings) do not influence overall service quality when customers have experienced any trouble during or after the online transaction. 6.1 Managerial implications The results of this study have many implications for online service retailers. The use of the two scales may improve the understanding of which dimensions are more relevant when the service is delivered properly (e-service quality scale) or when a service failure has taken place (e-service recovery quality scale). Managers can use these scales to measure and improve their service offerings, identifying strong and weak points of their e-service quality. In particular, our results reveal that online service companies need to place special emphasis on functionality and information. Accordingly, a practitioner could improve customers’ e-service quality perceptions by making the customer-website interaction a pleasant and trouble-free experience (e.g. interacting with the site has to be easy, the site has to be well organised, intuitive and connected to other relevant sites, which are useful). In addition, managers could benefit from appropriate content management systems. Managers should not publish content just because they have it, but focus on what is relevant to their customers (e.g. full details of service characteristics and prices). Additionally, the language used on the website should not confuse customers. Writing on a direct and concise manner would improve information quality. The fact that reliability/fulfilment and customer recovery did not have a significant effect on e-service quality has interesting implications for management. It seems that in our study companies are not handling customers’ problems very effectively. In the online context (where many alternatives are only a mouse click away), firms can not ignore customers when service failures occur. As such, an important priority for managers is to assign the resources needed to provide a successful service recovery. Indeed, a correct eservice recovery can foster customers’ positive reactions such as giving a second chance to the firm, or creating positive word of mouth, and could avoid customers’ negative reactions (e.g. negative word of mouth and service provider switching). 6.2 Limitations and issues for further research The e-service quality area is in an early stage of research and numerous possibilities still exist for exploring and expanding the knowledge in this area. Future qualitative research might need to be undertaken to further explain the influence of post-services (customer 168 E. Fernández Sabiote and S. Román recovery) not only on overall e-service quality, but also on customers’ satisfaction, trust, loyalty or word of mouth. Additionally, there is growing evidence of customer frustration in self-service experiences. However, it has been overlooked how e-service quality dimensions can affect customers’ frustration and how this frustration can influence customers’ future behaviours. It is likely that when a customer experiences a low level of functionality, the level of frustration may be totally different than when the customer experiences problems with privacy/security. Acknowledgements This study has received funding from the project ECO-2009-13170. References Akinci, S., Atilgan-Inan, E. and Aksoy, S. (2010) ‘Re-assessment of E-S-qual and E-RecS-qual in a pure service setting’, Journal of Business Research, Vol. 63, pp.232–240. Anderson, R.E. and Srinivasan, S.S. (2003) ‘E-satisfaction and E-loyalty: a contingency framework’, Psychology & Marketing, Vol. 20, No. 2, pp.123–138. Bagozzi, R. (1994), ‘Structural equation models in marketing research: basic principles’, in Bagozzi, R. (Ed.): Principles of Marketing Research, Blackwell, Oxford, pp.317–385. Barnes, S.J. and Vidgen, R.T. (2001) ‘An evaluation of cyber book-shops: the webqual method’, International Journal of Electronic Commerce, Vol. 6, No. 1, pp.11–30. Barnes, S.J. and Vidgen, R.T. (2004), ‘Interactive e-government: evaluating the web site of the UK inland revenue’, Journal of Electronic Commerce in Organizations, Vol. 2, No. 1, pp.42–62. Bauer, H.H., Falk, T. and Hammerschmidt, M. (2006), ‘eTransQual: a transaction process-based approach for capturing service quality in online shopping’, Journal of Business Research, Vol. 59, No. 7, pp.866–875. Brady, M.K. and Cronin, J.J.J. (2001) ‘Some new thoughts on conceptualizing perceived service quality: a hierarchical approach’, Journal of Marketing, Vol. 65, No. 3, pp.34–49. Cassel, C.M., Hackl, P. and Westlund, A.H. (2000), ‘On measurement of intangible assets: a study of fobustness of partial least squares’, Total Quality Management, Vol. 11, No. 7, pp.S897–S907. Chin, W.W. (1998) ‘The partial least squares approach to structural equation modeling’, in Marcoulides, G.A. (Ed.): Modern Methods for Business Research, Lawrence Erlbaum Associates, London, pp.295–358. Collier, J.E. and Bienstock, C.C. (2006) ‘Measuring service quality in e-retailing’, Journal of Service Research, Vol. 8, No. 3, pp.260–275. Collier, J.E. and Bienstock, C.C. (2009) ‘Model misspecification: contrasting formative and reflective indicators for a model of e-service quality’, Journal of Marketing Theory and Practice, Vol. 17, No. 3, pp.283–293. Cronin, J.J. (2003) ‘Looking back to see forward in services marketing: some ideas to consider’, Managing Service Quality, Vol. 13, No. 5, pp.332–337. Dabholkar, P.A., Thorpe, D.I. and Rentz, J.O. (1996) ‘A measure of service quality for retail stores’, Journal of Academy of Marketing Science, Vol. 24, No. 1, pp.3–16. Diamantopoulos, A. (2006) ‘The error term in formative measurement models: interpretation and modeling implications’, Journal of Modeling in Management, Vol. 1, No. 1, pp.7–17. Measuring e-service quality in the context of service sites 169 Diamantopoulos, A. and Winklhofer, H.M. (2001) ‘Index construction with formative indicators: an alternative to scale development’, Journal of Marketing Research, Vol. 38, No. 2, pp.269– 277. Fornell, C. and Cha, J. (1994) ‘Partial least squares’, in Bagozzi, R.P. (Ed.): Advanced Methods of Marketing Research, Basil Blackwell, Cambridge, MA, pp.52–78. Frambach, R.T., Roest, H.C.A. and Krishnan, T.V. (2007) ‘The impact of consumer internet experience on channel preference and usage intentions across the different stages of the buying process’, Journal of Interactive Marketing, Vol. 21, No. 2, pp.26–41. Francis, J.E. and White, L. (2002) ‘PIRQUAL: a scale for measuring customer expectations and perceptions of quality in internet retailing’, in Evans, K. and Scheer, L. (Eds): Marketing Educators’ Conference: Marketing Theory and Applications, Vol. 13, pp.263–270. Geiser, S. (1975) ‘The predictive samples reuse method with applications’, Journal of the American Statistical Association, Vol. 70, pp.320–328. Gounaris, S. and Dimitriadis, S. (2003) ‘Assessing service quality on the web: evidence from business-to-consumer portals’, Journal of Services Marketing, Vol. 17, No. 5, pp.529–548. Harris, L.C. and Goode, M.M.H. (2004) ‘The four levels of loyalty and the pivotal role of trust: a study of online service dynamics’, Journal of Retailing, Vol. 80, No. 2, pp.139–158. Hennig-Thurau, T. and Klee, A. (1997) ‘The impact of customer satisfaction and relationship quality on customer retention: a critical reassessment and model development’, Psychology & Marketing, Vol. 14, No. 8, pp.737–764. Jarvis, C.B., MacKenzie, S.B. and Podsakoff, P.M. (2003) ‘A critical review of construct indicators and measurement model misspecification in marketing and consumer research’, Journal of Consumer Research, Vol. 30, No. 2, pp.199–218. Kau, A.K., Tang Y.E. and Ghose S. (2003) ‘Typology of online shoppers’, Journal of Consumer Marketing, Vol. 20, No. 2, pp.139–156. Keen, C.A., Wetzels, M.B., De Ruyter, K. and Feinberg, D.R. (2004) ‘E-tailers versus retailers, which factors determine consumer preferences’, Journal of Business Research, Vol. 57, No. 7, pp.685–695. Kim, M., Kim, J-H. and Lennon, S.J. (2006) ‘Online service attributes available on apparel retail web sites: an E-S-QUAL approach’, Managing Service Quality, Vol. 16, No. 1, pp.51–77. Kim, E. and Lee, F. (2009) ‘E-service quality competition through personalization under consumer privacy concerns’, Electronic Commerce Research and Applications, Vol. 8, No. 4, pp.182– 190. Mackenzie, S.B., Podsakoff, P.M. and Jarvis, C.B. (2005) ‘The problem of measurement model misspecification in behavioral and organizational research and some recommended solutions’, Journal of Applied Psychology, Vol. 90, No. 4, pp.710–730. Maxham, J.G. and Netemeyer, R.G. (2002) ‘Modeling customer perceptions of complaint handling over time: the effects of perceived justice on satisfaction and intent’, Journal of Retailing, Vol. 78, No. 4, pp.239–252. Maxham, J.G. and Netemeyer, R.G. (2003), ‘Firms reap what they sow: the effects of shared values and perceived organizational justice on customers’ evaluations of complaint handling’, Journal of Marketing, Vol. 67, No. 1, pp.46–62. McGovern, G. and Norton, R. (2002) Content Critical: Gaining Competitive Adventage through High-Quality Web Content, Pearson Education, Great Britain. Montoya-Weiss, M.M., Voss, G.B. and Grewal, D. (2003) ‘Determinants of online channel use and overall satisfaction with a relational, multichannel service provider’, Journal of the Academy of Marketing Science, Vol. 31, No. 4, pp.448–458. Parasuraman, A., Zeithaml, V. and Berry, L.A. (1985) ‘Conceptual model of service quality and its implications for future research’, Journal of Marketing, Vol. 49, No. 4, pp.41–50. Parasuraman, A., Zeithaml, V. and Berry, L. (1988) ‘SERVQUAL: a multiple-item scale for measuring consumer perceptions of service quality’, Journal of Retailing, Vol. 64, No. 1, pp.12–40. 170 E. Fernández Sabiote and S. Román Parasuraman, A., Zeithaml, V. and Malhotra A. (2005) ‘E-S-QUAL a multiple-item sacel for assessing electronic service quality’, Journal of Service Research, Vol. 7, No. 3, pp.213–233. Rossiter, J.R. (2007) ‘Toward a valid measure of e-retailing service quality’, Journal of Theoretical and Applied Electronic Commerce Research, Vol. 2, No. 3, pp.36–48. Rowley, J. (2006) ‘An analysis of the e-service literature: towards a research agenda’, Internet Research, Vol. 16, No. 3, pp.339–359. Rust, R.T. and Lemon, K.N. (2001) ‘E-servie and the consumer’, International Journal of Electronic Commerce, Vol. 5, No. 3, pp.85–101. Sekaran, U. (2002) Research Methods for Business, 4th ed., Wiley, New York, NY. Stone, M. (1974) ‘Cross-validiation choice and assessment of statistical predictions’, Journal of the Royal Statistical Society, Vol. 36, No. 2, pp.111–133. Swinyard, W.R. and Smith, S.M. (2003) ‘Why people (don’t) shop online: a lifestyle study of the internet consumer’, Psychology & Marketing, Vol. 20, No. 7, pp.567–598. Tenenhaus, M., Esposito Vinzi, V., Chatelin, Y.M. and Lauro, C. (2005) ‘PLS path modeling’, Computational Statistics & Data Analysis, Vol. 48, No. 1, pp.159–205. Thompson, C.J., Locander, W.B. and Pollio, H.R. (1989) ‘Putting the consumer experience back into consumer research: the philosophy and methodo of existential-phenomenology’, Journal of Consumer Research, Vol. 16, No. 1, pp.133–147. White, J.C., Varadarajan, P.R. and Dacin, P.A. (2003) ‘Market situation interpretation and response: the role of cognitive style, organizational culture, and information use’, Journal of Marketing, Vol. 67, No. 3, pp.63–79. Wolfinbarger, M. and Gilly, M.C. (2003) ‘eTailQ: dimensionalizing, measuring and predicting etail quality’, Journal of Retailing, Vol. 79, No. 3, pp.183–198. Yang, Z., Cai, S., Zhou, Z. and Zhou N. (2005) ‘Development and validation of an instrument to measure user perceived service quality of information presenting web portals’, Information & Management, Vol. 42, No. 4, pp.575–589. Yang, Z., Jun, M. and Peterson, R.T. (2004) ‘Measuring customer perceived online service quality: scale development and managerial implications’, International Journal of Operations & Production Management, Vol. 24, Nos. 11/12, pp.1149–1174. Yang, Z., Peterson, R.T. and Cai, S. (2003) ‘Services quality dimensions of internet retailing: an exploratory analysis’, Journal of Service Marketing, Vol. 17, Nos. 6/7, pp.685–700. Zeithaml, V.A., Parasuraman, A. and Malhotra, A. (2000) E-service Quality: Definition, Dimensions and Conceptual Model, Working Paper, Marketing Science Institute, Cambridge, MA. Zeithaml, V.A., Parasuraman, A. and Malhotra A. (2002) ‘Service quality delivery through web sites: a critical review of extant knowledge’, Journal of the Academy of Marketing Science, Vol. 30, No. 4, pp.362–375. Notes 1 2 3 4 In particular, results from MacKenzie et al. (2005) revealed that model misspecification can inflate unstandardised structural parameter estimates by as much as 400% or deflate them by as much as 80%. Note that most of these items express similar ideas. The first focus group started from 51 items. They suggested to eliminate 8, reword 7 and add one item. In the same manner, the second focus group, which started from 44, suggested to eliminate 6, reword 6 and add five items. The official website of this University discloses both, services and faculty staff e-mails addresses. Measuring e-service quality in the context of service sites 5 6 7 171 A total of 32 e-mails were returned as undeliverable messages. There were 567 responses (43% response rate). Of these, 368 replied they did not meet the conditions to qualify for the survey and 199 completed the questionnaire. PLS analysis ensures optimal prediction accuracy because its estimates have been shown to be very robust against multi-collinearity (Tenenhaus et al., 2005). 89 subjects contacted, either by phone or email, the service provider, during or after the purchase. Appendix A Measurement items Functionality Information at this site is well organised. The number and type of links are meaningful. This site makes it easy to find what I need. The registration process is simple. This site does not crash. Information The contents in the website are easy to understand. Full details of service characteristics are available. Full details of service pricing are available. Privacy/security This website clearly informs about the use of registered data. The website only asks for data necessary to make the online transaction. Security features at this site are easy to understand. Reliability/fulfilment It is truthful about its offerings. What is offered in this site is really available. Promises to do something by a certain time, they do it. Customer Recovery I can easily contact a customer service representative over the telephone. Inquires and complaints are solved correctly. This site gives me personalised attention. Overall e-service quality Overall, the services provided by the portal have excellent quality. This portal’s service offering are very competitive. The service quality provided by this portal is excellent. 172 E. Fernández Sabiote and S. Román Satisfaction My choice to purchase from this website was a wise one. I am satisfied with my decision to purchase from this website. I think I did the right thing by buying from this website. Trust I feel confident at using this website. This website is reliable for online shopping. I can trust the performance of this website to be good.